mcpField
R
remember

remember

github.com/parcadeiAgent workflows3.9k523 installsUpdated 40w ago
Repository

About remember

remember is a skill for saving learnings, patterns, and decisions to a memory system for later recall. It stores entries in PostgreSQL with BGE embeddings and can infer a learning type and extract tags from the content.

Key features

  1. 01Store learnings in PostgreSQL
  2. 02Classify entries by learning type
  3. 03Automatically infer entry type
  4. 04Extract tags from content
  5. 05Return a confirmation ID

Use cases

  1. 01Save a working solution for later
  2. 02Record an architectural decision
  3. 03Keep a discovered codebase pattern
  4. 04Document a failed approach or error fix

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/parcadei/continuous-claude-v3 --skill remember

After installing, type/rememberin Claude Code to confirm it loaded.

Details

From the project README

Remember - Store Learning in Memory

Store a learning, pattern, or decision in the memory system for future recall.

Usage

/remember <what you learned>

Or with explicit type:

/remember --type WORKING_SOLUTION <what you learned>

Examples

/remember TypeScript hooks require npm install before they work
/remember --type ARCHITECTURAL_DECISION Session affinity uses terminal PID
/remember --type FAILED_APPROACH Don't use subshell for store_learning command

What It Does

  1. Stores the learning in PostgreSQL with BGE embeddings
  2. Auto-detects learning type if not specified
  3. Extracts tags from content
  4. Returns confirmation with ID

Learning Types

TypeUse For
WORKING_SOLUTIONFixes, solutions that worked (default)
ARCHITECTURAL_DECISIONDesign choices, system structure
CODEBASE_PATTERNPatterns discovered in code
FAILED_APPROACHWhat didn't work
ERROR_FIXSpecific error resolutions

Execution

When this skill is invoked, run:

cd $CLAUDE_OPC_DIR && PYTHONPATH=. uv run python scripts/core/store_learning.py \
  --session-id "manual-$(date +%Y%m%d-%H%M)" \
  --type <TYPE or WORKING_SOLUTION> \
  --content "<ARGS>" \
  --context "manual entry via /remember" \
  --confidence medium

Auto-Type Detection

If no --type specified, infer from content:

  • Contains "error", "fix", "bug" → ERROR_FIX
  • Contains "decided", "chose", "architecture" → ARCHITECTURAL_DECISION
  • Contains "pattern", "always", "convention" → CODEBASE_PATTERN
  • Contains "failed", "didn't work", "don't" → FAILED_APPROACH
  • Default → WORKING_SOLUTION

Comments

Markdown supported